diff --git a/plwn-api_plwn_dump_new_07-12-2022.sqlite b/plwn-api_plwn_dump_new_07-12-2022.sqlite
new file mode 100644
index 0000000000000000000000000000000000000000..19fe03e3c05a58941aea3e5207be305382858d67
Binary files /dev/null and b/plwn-api_plwn_dump_new_07-12-2022.sqlite differ
diff --git a/plwn/bases.py b/plwn/bases.py
index 0e66bc704b21000b86afe8068fdb41a16d3d6926..09e84e3da0d283c463ebae82412e0a812305f245 100644
--- a/plwn/bases.py
+++ b/plwn/bases.py
@@ -588,7 +588,7 @@ class SynsetBase(object):
:meth:`LexicalUnitBase.to_dict`.
"""
syn_dict = {
- u'id': self.id,
+ u'id': self.uuid,
u'definition': self.definition,
u'is_artificial': self.is_artificial,
u'units': tuple(
@@ -602,7 +602,7 @@ class SynsetBase(object):
if include_related:
syn_dict[u'related'] = {
six.text_type(rel): tuple(
- (target.id, target.short_str())
+ (target.uuid, target.short_str())
for target in self.related(rel)
)
for rel in self.relations
@@ -631,7 +631,7 @@ class SynsetBase(object):
def __repr__(self):
head = self.lexical_units[0]
rstr = '<Synset id={!r} lemma={!r} pos={!r} variant={!r}'.format(
- self.id,
+ str(self.uuid),
head.lemma,
head.pos,
head.variant,
@@ -938,7 +938,7 @@ class LexicalUnitBase(object):
}
"""
lu_dict = {
- u'id': self.id,
+ u'id': self.uuid,
u'legacy_id': self.legacy_id,
u'lemma': self.lemma,
u'pos': self.pos.value,
@@ -949,7 +949,7 @@ class LexicalUnitBase(object):
u'external_links': tuple(self.external_links),
u'usage_notes': tuple(self.usage_notes),
u'domain': self.domain.value,
- u'synset': self.synset.id,
+ u'synset': self.synset.uuid,
u'verb_aspect': None
if self.verb_aspect is None
else self.verb_aspect.value,
@@ -966,7 +966,7 @@ class LexicalUnitBase(object):
if include_related:
lu_dict[u'related'] = {
six.text_type(rel): tuple(
- (target.id, six.text_type(target))
+ (target.uuid, six.text_type(target))
for target in self.related(rel)
)
for rel in self.relations
@@ -1002,7 +1002,7 @@ class LexicalUnitBase(object):
def __repr__(self):
return '<LexicalUnit id={!r} lemma={!r} pos={!r} variant={!r}>'.format(
- self.id,
+ str(self.uuid),
self.lemma,
self.pos,
self.variant,
diff --git a/plwn/storages/sqlite.py b/plwn/storages/sqlite.py
index 940d4b11764bedbf70ef409771d257cd2f997de3..1e7ec64e646256706640296a6aaa2904cb7d47b3 100644
--- a/plwn/storages/sqlite.py
+++ b/plwn/storages/sqlite.py
@@ -102,7 +102,7 @@ CREATE UNIQUE INDEX value ON tbl_domain (value);
-- Synset only gets one simple table
CREATE TABLE tbl_synset (
- id BLOB PRIMARY KEY,
+ id INTEGER PRIMARY KEY,
legacy_id INTEGER NULL ,
definition TEXT COLLATE locale,
isartificial INTEGER NOT NULL DEFAULT 0
@@ -112,13 +112,13 @@ CREATE TABLE tbl_synset (
-- properties. They also need indexes for lookup.
CREATE TABLE tbl_lexicalunit (
- id BLOB PRIMARY KEY,
+ id INTEGER PRIMARY KEY,
legacy_id INTEGER NULL ,
lemma TEXT NOT NULL COLLATE locale,
pos INTEGER NOT NULL
REFERENCES tbl_pos (id),
variant INTEGER NOT NULL ,
- synset BLOB NOT NULL
+ synset INTEGER NOT NULL
REFERENCES tbl_synset (id),
unitindex INTEGER NOT NULL ,
definition TEXT COLLATE locale,
@@ -139,7 +139,7 @@ CREATE UNIQUE INDEX synset ON tbl_lexicalunit (synset, unitindex);
-- Tables dependant on lexicalunit
CREATE TABLE tbl_senseexample (
- unitid BLOB NOT NULL
+ unitid INTEGER NOT NULL
REFERENCES tbl_lexicalunit (id),
example TEXT NOT NULL COLLATE locale,
source TEXT NOT NULL COLLATE locale
@@ -147,21 +147,21 @@ CREATE TABLE tbl_senseexample (
CREATE INDEX sen_i ON tbl_senseexample (unitid);
CREATE TABLE tbl_externallink (
- unitid BLOB NOT NULL
+ unitid INTEGER NOT NULL
REFERENCES tbl_lexicalunit (id),
link TEXT NOT NULL COLLATE locale
);
CREATE INDEX link_i ON tbl_externallink (unitid);
CREATE TABLE tbl_usagenote (
- unitid BLOB NOT NULL
+ unitid INTEGER NOT NULL
REFERENCES tbl_lexicalunit (id),
note TEXT NOT NULL COLLATE locale
);
CREATE INDEX note_i ON tbl_usagenote (unitid);
CREATE TABLE tbl_unitemotionname (
- unitid BLOB NOT NULL
+ unitid INTEGER NOT NULL
REFERENCES tbl_lexicalunit (id),
nameid INTEGER NOT NULL
REFERENCES tbl_emotionname (id),
@@ -169,7 +169,7 @@ CREATE TABLE tbl_unitemotionname (
);
CREATE TABLE tbl_unitemotionvaluation (
- unitid BLOB NOT NULL
+ unitid INTEGER NOT NULL
REFERENCES tbl_lexicalunit (id),
valuationid INTEGER NOT NULL
REFERENCES tbl_emotionvaluation (id),
@@ -229,7 +229,7 @@ CREATE UNIQUE INDEX parentpart ON tbl_lexicalrelationtype (
-- Reverse indexes are needed, too.
CREATE TABLE tbl_synsetrelationalias (
name TEXT PRIMARY KEY NOT NULL COLLATE locale,
- relationid BLOB NOT NULL
+ relationid INTEGER NOT NULL
REFERENCES tbl_synsetrelationtype (id)
);
CREATE INDEX synsetrelationalias_irev ON tbl_synsetrelationalias (
@@ -237,7 +237,7 @@ CREATE INDEX synsetrelationalias_irev ON tbl_synsetrelationalias (
);
CREATE TABLE tbl_lexicalrelationalias (
name TEXT PRIMARY KEY NOT NULL COLLATE locale,
- relationid BLOB NOT NULL
+ relationid INTEGER NOT NULL
REFERENCES tbl_lexicalrelationtype (id)
);
CREATE INDEX lexicalrelationalias_irev ON tbl_lexicalrelationalias (
@@ -246,21 +246,21 @@ CREATE INDEX lexicalrelationalias_irev ON tbl_lexicalrelationalias (
-- Next are finally the relation instances
CREATE TABLE tbl_synsetrelation (
- source BLOB NOT NULL
+ source INTEGER NOT NULL
REFERENCES tbl_synset (id),
- relationtype BLOB NOT NULL
+ relationtype INTEGER NOT NULL
REFERENCES tbl_synsetrelationtype (id),
- target BLOB NOT NULL
+ target INTEGER NOT NULL
REFERENCES tbl_synset (id),
PRIMARY KEY (source, relationtype, target)
);
CREATE TABLE tbl_lexicalrelation (
- source BLOB NOT NULL
+ source INTEGER NOT NULL
REFERENCES tbl_lexicalunit (id),
- relationtype BLOB NOT NULL
+ relationtype INTEGER NOT NULL
REFERENCES tbl_lexicalrelationtype (id),
- target BLOB NOT NULL
+ target INTEGER NOT NULL
REFERENCES tbl_lexicalunit (id),
PRIMARY KEY (source, relationtype, target)
@@ -694,6 +694,7 @@ class LexicalUnit(bs.LexicalUnitBase):
self._var = variant
self._synid = synid
# Rest is unitialized
+ self._uuid = _UNFETCHED
self._leg_id = _UNFETCHED
self._syn = _UNFETCHED
self._def = _UNFETCHED
@@ -745,6 +746,17 @@ class LexicalUnit(bs.LexicalUnitBase):
self._leg_id = cur.fetchone()[0]
return self._leg_id
+ @property
+ def uuid(self):
+ if self._uuid is _UNFETCHED:
+ with closing(self._db.cursor()) as cur:
+ cur.execute(
+ u"SELECT quote(id) from tbl_lexicalunit WHERE id = ?",
+ (self._id,),
+ )
+ self._uuid = cur.fetchone()[0][2:-1]
+ return self._uuid
+
@property
def synset(self):
if self._syn is _UNFETCHED or self._syn() is None:
@@ -1024,6 +1036,7 @@ class Synset(bs.SynsetBase):
self._id = syn_id
self._isart = syn_art
+ self._uuid = _UNFETCHED
self._units = _UNFETCHED
self._def = _UNFETCHED
self._leg_id = _UNFETCHED
@@ -1096,6 +1109,17 @@ class Synset(bs.SynsetBase):
self._leg_id = cur.fetchone()[0]
return self._leg_id
+ @property
+ def uuid(self):
+ if self._uuid is _UNFETCHED:
+ with closing(self._db.cursor()) as cur:
+ cur.execute(
+ u"SELECT quote(id) from tbl_synset WHERE id = ?",
+ (self._id,),
+ )
+ self._uuid = cur.fetchone()[0][2:-1]
+ return self._uuid
+
@property
def definition(self):
if self._def is _UNFETCHED:
@@ -1408,78 +1432,6 @@ class _DBBuilder(object):
en.RelationKind.lexical,
)
- def _show(self):
- with closing(self._db.cursor()) as cur:
- cur.execute(
- u"""
- SELECT * FROM tbl_synset
- """
- )
- print(cur.fetchall())
- with closing(self._db.cursor()) as cur:
- cur.execute(
- u"""
- SELECT * FROM tbl_synsetrelation
- """
- )
- print(cur.fetchall())
- with closing(self._db.cursor()) as cur:
- cur.execute(
- u"""
- SELECT * FROM tbl_synsetrelationtype
- """
- )
- print(cur.fetchall())
- with closing(self._db.cursor()) as cur:
- cur.execute(
- u"""
- SELECT * FROM tbl_synsetrelationparentpart
- """
- )
- print(cur.fetchall())
- with closing(self._db.cursor()) as cur:
- cur.execute(
- u"""
- SELECT * FROM tbl_synsetrelationchildpart
- """
- )
- print(cur.fetchall())
- with closing(self._db.cursor()) as cur:
- cur.execute(
- u"""
- SELECT * FROM tbl_lexicalunit
- """
- )
- print(cur.fetchall())
- with closing(self._db.cursor()) as cur:
- cur.execute(
- u"""
- SELECT * FROM tbl_lexicalrelation
- """
- )
- print(cur.fetchall())
- with closing(self._db.cursor()) as cur:
- cur.execute(
- u"""
- SELECT * FROM tbl_lexicalrelationtype
- """
- )
- print(cur.fetchall())
- with closing(self._db.cursor()) as cur:
- cur.execute(
- u"""
- SELECT * FROM tbl_lexicalrelationparentpart
- """
- )
- print(cur.fetchall())
- with closing(self._db.cursor()) as cur:
- cur.execute(
- u"""
- SELECT * FROM tbl_lexicalrelationchildpart
- """
- )
- print(cur.fetchall())
-
def _insert_synset(self, syn_node):
self._db.execute(
u"""